1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227222822292230223122322233223422352236223722382239224022412242224322442245224622472248224922502251225222532254225522562257225822592260226122622263226422652266226722682269227022712272227322742275227622772278227922802281228222832284228522862287228822892290229122922293229422952296229722982299230023012302230323042305230623072308230923102311231223132314231523162317231823192320232123222323232423252326232723282329233023312332233323342335233623372338233923402341234223432344234523462347234823492350235123522353235423552356235723582359236023612362236323642365236623672368236923702371237223732374237523762377237823792380238123822383238423852386238723882389239023912392239323942395239623972398239924002401240224032404240524062407240824092410241124122413241424152416241724182419242024212422242324242425242624272428242924302431243224332434243524362437243824392440244124422443244424452446244724482449245024512452245324542455245624572458245924602461246224632464246524662467246824692470247124722473247424752476247724782479248024812482248324842485248624872488248924902491249224932494249524962497249824992500250125022503250425052506250725082509251025112512251325142515251625172518251925202521252225232524252525262527252825292530253125322533253425352536253725382539254025412542254325442545254625472548254925502551255225532554255525562557255825592560256125622563256425652566256725682569257025712572257325742575257625772578257925802581258225832584258525862587258825892590259125922593259425952596259725982599260026012602260326042605260626072608260926102611261226132614261526162617261826192620262126222623262426252626262726282629263026312632263326342635263626372638263926402641264226432644264526462647264826492650265126522653265426552656265726582659266026612662266326642665266626672668266926702671267226732674267526762677267826792680268126822683268426852686268726882689269026912692269326942695269626972698269927002701270227032704270527062707270827092710271127122713271427152716271727182719272027212722272327242725272627272728272927302731273227332734273527362737273827392740274127422743274427452746274727482749275027512752275327542755275627572758275927602761276227632764276527662767276827692770277127722773277427752776277727782779278027812782278327842785278627872788278927902791279227932794279527962797279827992800280128022803280428052806280728082809281028112812281328142815281628172818281928202821282228232824282528262827282828292830283128322833283428352836283728382839284028412842284328442845284628472848284928502851285228532854285528562857285828592860286128622863286428652866286728682869287028712872287328742875287628772878287928802881288228832884288528862887288828892890289128922893289428952896289728982899290029012902290329042905290629072908 |
- "use strict";
- Object.defineProperty(exports, "__esModule", { value: true });
- exports.Tokenizer = exports.TokenizerMode = void 0;
- const preprocessor_js_1 = require("./preprocessor.js");
- const unicode_js_1 = require("../common/unicode.js");
- const token_js_1 = require("../common/token.js");
- const decode_js_1 = require("entities/lib/decode.js");
- const error_codes_js_1 = require("../common/error-codes.js");
- const html_js_1 = require("../common/html.js");
- //C1 Unicode control character reference replacements
- const C1_CONTROLS_REFERENCE_REPLACEMENTS = new Map([
- [0x80, 8364],
- [0x82, 8218],
- [0x83, 402],
- [0x84, 8222],
- [0x85, 8230],
- [0x86, 8224],
- [0x87, 8225],
- [0x88, 710],
- [0x89, 8240],
- [0x8a, 352],
- [0x8b, 8249],
- [0x8c, 338],
- [0x8e, 381],
- [0x91, 8216],
- [0x92, 8217],
- [0x93, 8220],
- [0x94, 8221],
- [0x95, 8226],
- [0x96, 8211],
- [0x97, 8212],
- [0x98, 732],
- [0x99, 8482],
- [0x9a, 353],
- [0x9b, 8250],
- [0x9c, 339],
- [0x9e, 382],
- [0x9f, 376],
- ]);
- //States
- var State;
- (function (State) {
- State[State["DATA"] = 0] = "DATA";
- State[State["RCDATA"] = 1] = "RCDATA";
- State[State["RAWTEXT"] = 2] = "RAWTEXT";
- State[State["SCRIPT_DATA"] = 3] = "SCRIPT_DATA";
- State[State["PLAINTEXT"] = 4] = "PLAINTEXT";
- State[State["TAG_OPEN"] = 5] = "TAG_OPEN";
- State[State["END_TAG_OPEN"] = 6] = "END_TAG_OPEN";
- State[State["TAG_NAME"] = 7] = "TAG_NAME";
- State[State["RCDATA_LESS_THAN_SIGN"] = 8] = "RCDATA_LESS_THAN_SIGN";
- State[State["RCDATA_END_TAG_OPEN"] = 9] = "RCDATA_END_TAG_OPEN";
- State[State["RCDATA_END_TAG_NAME"] = 10] = "RCDATA_END_TAG_NAME";
- State[State["RAWTEXT_LESS_THAN_SIGN"] = 11] = "RAWTEXT_LESS_THAN_SIGN";
- State[State["RAWTEXT_END_TAG_OPEN"] = 12] = "RAWTEXT_END_TAG_OPEN";
- State[State["RAWTEXT_END_TAG_NAME"] = 13] = "RAWTEXT_END_TAG_NAME";
- State[State["SCRIPT_DATA_LESS_THAN_SIGN"] = 14] = "SCRIPT_DATA_LESS_THAN_SIGN";
- State[State["SCRIPT_DATA_END_TAG_OPEN"] = 15] = "SCRIPT_DATA_END_TAG_OPEN";
- State[State["SCRIPT_DATA_END_TAG_NAME"] = 16] = "SCRIPT_DATA_END_TAG_NAME";
- State[State["SCRIPT_DATA_ESCAPE_START"] = 17] = "SCRIPT_DATA_ESCAPE_START";
- State[State["SCRIPT_DATA_ESCAPE_START_DASH"] = 18] = "SCRIPT_DATA_ESCAPE_START_DASH";
- State[State["SCRIPT_DATA_ESCAPED"] = 19] = "SCRIPT_DATA_ESCAPED";
- State[State["SCRIPT_DATA_ESCAPED_DASH"] = 20] = "SCRIPT_DATA_ESCAPED_DASH";
- State[State["SCRIPT_DATA_ESCAPED_DASH_DASH"] = 21] = "SCRIPT_DATA_ESCAPED_DASH_DASH";
- State[State["SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN"] = 22] = "SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN";
- State[State["SCRIPT_DATA_ESCAPED_END_TAG_OPEN"] = 23] = "SCRIPT_DATA_ESCAPED_END_TAG_OPEN";
- State[State["SCRIPT_DATA_ESCAPED_END_TAG_NAME"] = 24] = "SCRIPT_DATA_ESCAPED_END_TAG_NAME";
- State[State["SCRIPT_DATA_DOUBLE_ESCAPE_START"] = 25] = "SCRIPT_DATA_DOUBLE_ESCAPE_START";
- State[State["SCRIPT_DATA_DOUBLE_ESCAPED"] = 26] = "SCRIPT_DATA_DOUBLE_ESCAPED";
- State[State["SCRIPT_DATA_DOUBLE_ESCAPED_DASH"] = 27] = "SCRIPT_DATA_DOUBLE_ESCAPED_DASH";
- State[State["SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH"] = 28] = "SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH";
- State[State["SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN"] = 29] = "SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN";
- State[State["SCRIPT_DATA_DOUBLE_ESCAPE_END"] = 30] = "SCRIPT_DATA_DOUBLE_ESCAPE_END";
- State[State["BEFORE_ATTRIBUTE_NAME"] = 31] = "BEFORE_ATTRIBUTE_NAME";
- State[State["ATTRIBUTE_NAME"] = 32] = "ATTRIBUTE_NAME";
- State[State["AFTER_ATTRIBUTE_NAME"] = 33] = "AFTER_ATTRIBUTE_NAME";
- State[State["BEFORE_ATTRIBUTE_VALUE"] = 34] = "BEFORE_ATTRIBUTE_VALUE";
- State[State["ATTRIBUTE_VALUE_DOUBLE_QUOTED"] = 35] = "ATTRIBUTE_VALUE_DOUBLE_QUOTED";
- State[State["ATTRIBUTE_VALUE_SINGLE_QUOTED"] = 36] = "ATTRIBUTE_VALUE_SINGLE_QUOTED";
- State[State["ATTRIBUTE_VALUE_UNQUOTED"] = 37] = "ATTRIBUTE_VALUE_UNQUOTED";
- State[State["AFTER_ATTRIBUTE_VALUE_QUOTED"] = 38] = "AFTER_ATTRIBUTE_VALUE_QUOTED";
- State[State["SELF_CLOSING_START_TAG"] = 39] = "SELF_CLOSING_START_TAG";
- State[State["BOGUS_COMMENT"] = 40] = "BOGUS_COMMENT";
- State[State["MARKUP_DECLARATION_OPEN"] = 41] = "MARKUP_DECLARATION_OPEN";
- State[State["COMMENT_START"] = 42] = "COMMENT_START";
- State[State["COMMENT_START_DASH"] = 43] = "COMMENT_START_DASH";
- State[State["COMMENT"] = 44] = "COMMENT";
- State[State["COMMENT_LESS_THAN_SIGN"] = 45] = "COMMENT_LESS_THAN_SIGN";
- State[State["COMMENT_LESS_THAN_SIGN_BANG"] = 46] = "COMMENT_LESS_THAN_SIGN_BANG";
- State[State["COMMENT_LESS_THAN_SIGN_BANG_DASH"] = 47] = "COMMENT_LESS_THAN_SIGN_BANG_DASH";
- State[State["COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH"] = 48] = "COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH";
- State[State["COMMENT_END_DASH"] = 49] = "COMMENT_END_DASH";
- State[State["COMMENT_END"] = 50] = "COMMENT_END";
- State[State["COMMENT_END_BANG"] = 51] = "COMMENT_END_BANG";
- State[State["DOCTYPE"] = 52] = "DOCTYPE";
- State[State["BEFORE_DOCTYPE_NAME"] = 53] = "BEFORE_DOCTYPE_NAME";
- State[State["DOCTYPE_NAME"] = 54] = "DOCTYPE_NAME";
- State[State["AFTER_DOCTYPE_NAME"] = 55] = "AFTER_DOCTYPE_NAME";
- State[State["AFTER_DOCTYPE_PUBLIC_KEYWORD"] = 56] = "AFTER_DOCTYPE_PUBLIC_KEYWORD";
- State[State["BEFORE_DOCTYPE_PUBLIC_IDENTIFIER"] = 57] = "BEFORE_DOCTYPE_PUBLIC_IDENTIFIER";
- State[State["DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED"] = 58] = "DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED";
- State[State["DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED"] = 59] = "DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED";
- State[State["AFTER_DOCTYPE_PUBLIC_IDENTIFIER"] = 60] = "AFTER_DOCTYPE_PUBLIC_IDENTIFIER";
- State[State["BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS"] = 61] = "BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS";
- State[State["AFTER_DOCTYPE_SYSTEM_KEYWORD"] = 62] = "AFTER_DOCTYPE_SYSTEM_KEYWORD";
- State[State["BEFORE_DOCTYPE_SYSTEM_IDENTIFIER"] = 63] = "BEFORE_DOCTYPE_SYSTEM_IDENTIFIER";
- State[State["DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED"] = 64] = "DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED";
- State[State["DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED"] = 65] = "DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED";
- State[State["AFTER_DOCTYPE_SYSTEM_IDENTIFIER"] = 66] = "AFTER_DOCTYPE_SYSTEM_IDENTIFIER";
- State[State["BOGUS_DOCTYPE"] = 67] = "BOGUS_DOCTYPE";
- State[State["CDATA_SECTION"] = 68] = "CDATA_SECTION";
- State[State["CDATA_SECTION_BRACKET"] = 69] = "CDATA_SECTION_BRACKET";
- State[State["CDATA_SECTION_END"] = 70] = "CDATA_SECTION_END";
- State[State["CHARACTER_REFERENCE"] = 71] = "CHARACTER_REFERENCE";
- State[State["NAMED_CHARACTER_REFERENCE"] = 72] = "NAMED_CHARACTER_REFERENCE";
- State[State["AMBIGUOUS_AMPERSAND"] = 73] = "AMBIGUOUS_AMPERSAND";
- State[State["NUMERIC_CHARACTER_REFERENCE"] = 74] = "NUMERIC_CHARACTER_REFERENCE";
- State[State["HEXADEMICAL_CHARACTER_REFERENCE_START"] = 75] = "HEXADEMICAL_CHARACTER_REFERENCE_START";
- State[State["HEXADEMICAL_CHARACTER_REFERENCE"] = 76] = "HEXADEMICAL_CHARACTER_REFERENCE";
- State[State["DECIMAL_CHARACTER_REFERENCE"] = 77] = "DECIMAL_CHARACTER_REFERENCE";
- State[State["NUMERIC_CHARACTER_REFERENCE_END"] = 78] = "NUMERIC_CHARACTER_REFERENCE_END";
- })(State || (State = {}));
- //Tokenizer initial states for different modes
- exports.TokenizerMode = {
- DATA: State.DATA,
- RCDATA: State.RCDATA,
- RAWTEXT: State.RAWTEXT,
- SCRIPT_DATA: State.SCRIPT_DATA,
- PLAINTEXT: State.PLAINTEXT,
- CDATA_SECTION: State.CDATA_SECTION,
- };
- //Utils
- //OPTIMIZATION: these utility functions should not be moved out of this module. V8 Crankshaft will not inline
- //this functions if they will be situated in another module due to context switch.
- //Always perform inlining check before modifying this functions ('node --trace-inlining').
- function isAsciiDigit(cp) {
- return cp >= unicode_js_1.CODE_POINTS.DIGIT_0 && cp <= unicode_js_1.CODE_POINTS.DIGIT_9;
- }
- function isAsciiUpper(cp) {
- return cp >= unicode_js_1.CODE_POINTS.LATIN_CAPITAL_A && cp <= unicode_js_1.CODE_POINTS.LATIN_CAPITAL_Z;
- }
- function isAsciiLower(cp) {
- return cp >= unicode_js_1.CODE_POINTS.LATIN_SMALL_A && cp <= unicode_js_1.CODE_POINTS.LATIN_SMALL_Z;
- }
- function isAsciiLetter(cp) {
- return isAsciiLower(cp) || isAsciiUpper(cp);
- }
- function isAsciiAlphaNumeric(cp) {
- return isAsciiLetter(cp) || isAsciiDigit(cp);
- }
- function isAsciiUpperHexDigit(cp) {
- return cp >= unicode_js_1.CODE_POINTS.LATIN_CAPITAL_A && cp <= unicode_js_1.CODE_POINTS.LATIN_CAPITAL_F;
- }
- function isAsciiLowerHexDigit(cp) {
- return cp >= unicode_js_1.CODE_POINTS.LATIN_SMALL_A && cp <= unicode_js_1.CODE_POINTS.LATIN_SMALL_F;
- }
- function isAsciiHexDigit(cp) {
- return isAsciiDigit(cp) || isAsciiUpperHexDigit(cp) || isAsciiLowerHexDigit(cp);
- }
- function toAsciiLower(cp) {
- return cp + 32;
- }
- function isWhitespace(cp) {
- return cp === unicode_js_1.CODE_POINTS.SPACE || cp === unicode_js_1.CODE_POINTS.LINE_FEED || cp === unicode_js_1.CODE_POINTS.TABULATION || cp === unicode_js_1.CODE_POINTS.FORM_FEED;
- }
- function isEntityInAttributeInvalidEnd(nextCp) {
- return nextCp === unicode_js_1.CODE_POINTS.EQUALS_SIGN || isAsciiAlphaNumeric(nextCp);
- }
- function isScriptDataDoubleEscapeSequenceEnd(cp) {
- return isWhitespace(cp) || cp === unicode_js_1.CODE_POINTS.SOLIDUS || cp === unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN;
- }
- //Tokenizer
- class Tokenizer {
- constructor(options, handler) {
- this.options = options;
- this.handler = handler;
- this.paused = false;
- /** Ensures that the parsing loop isn't run multiple times at once. */
- this.inLoop = false;
- /**
- * Indicates that the current adjusted node exists, is not an element in the HTML namespace,
- * and that it is not an integration point for either MathML or HTML.
- *
- * @see {@link https://html.spec.whatwg.org/multipage/parsing.html#tree-construction}
- */
- this.inForeignNode = false;
- this.lastStartTagName = '';
- this.active = false;
- this.state = State.DATA;
- this.returnState = State.DATA;
- this.charRefCode = -1;
- this.consumedAfterSnapshot = -1;
- this.currentCharacterToken = null;
- this.currentToken = null;
- this.currentAttr = { name: '', value: '' };
- this.preprocessor = new preprocessor_js_1.Preprocessor(handler);
- this.currentLocation = this.getCurrentLocation(-1);
- }
- //Errors
- _err(code) {
- var _a, _b;
- (_b = (_a = this.handler).onParseError) === null || _b === void 0 ? void 0 : _b.call(_a, this.preprocessor.getError(code));
- }
- // NOTE: `offset` may never run across line boundaries.
- getCurrentLocation(offset) {
- if (!this.options.sourceCodeLocationInfo) {
- return null;
- }
- return {
- startLine: this.preprocessor.line,
- startCol: this.preprocessor.col - offset,
- startOffset: this.preprocessor.offset - offset,
- endLine: -1,
- endCol: -1,
- endOffset: -1,
- };
- }
- _runParsingLoop() {
- if (this.inLoop)
- return;
- this.inLoop = true;
- while (this.active && !this.paused) {
- this.consumedAfterSnapshot = 0;
- const cp = this._consume();
- if (!this._ensureHibernation()) {
- this._callState(cp);
- }
- }
- this.inLoop = false;
- }
- //API
- pause() {
- this.paused = true;
- }
- resume(writeCallback) {
- if (!this.paused) {
- throw new Error('Parser was already resumed');
- }
- this.paused = false;
- // Necessary for synchronous resume.
- if (this.inLoop)
- return;
- this._runParsingLoop();
- if (!this.paused) {
- writeCallback === null || writeCallback === void 0 ? void 0 : writeCallback();
- }
- }
- write(chunk, isLastChunk, writeCallback) {
- this.active = true;
- this.preprocessor.write(chunk, isLastChunk);
- this._runParsingLoop();
- if (!this.paused) {
- writeCallback === null || writeCallback === void 0 ? void 0 : writeCallback();
- }
- }
- insertHtmlAtCurrentPos(chunk) {
- this.active = true;
- this.preprocessor.insertHtmlAtCurrentPos(chunk);
- this._runParsingLoop();
- }
- //Hibernation
- _ensureHibernation() {
- if (this.preprocessor.endOfChunkHit) {
- this._unconsume(this.consumedAfterSnapshot);
- this.active = false;
- return true;
- }
- return false;
- }
- //Consumption
- _consume() {
- this.consumedAfterSnapshot++;
- return this.preprocessor.advance();
- }
- _unconsume(count) {
- this.consumedAfterSnapshot -= count;
- this.preprocessor.retreat(count);
- }
- _reconsumeInState(state, cp) {
- this.state = state;
- this._callState(cp);
- }
- _advanceBy(count) {
- this.consumedAfterSnapshot += count;
- for (let i = 0; i < count; i++) {
- this.preprocessor.advance();
- }
- }
- _consumeSequenceIfMatch(pattern, caseSensitive) {
- if (this.preprocessor.startsWith(pattern, caseSensitive)) {
- // We will already have consumed one character before calling this method.
- this._advanceBy(pattern.length - 1);
- return true;
- }
- return false;
- }
- //Token creation
- _createStartTagToken() {
- this.currentToken = {
- type: token_js_1.TokenType.START_TAG,
- tagName: '',
- tagID: html_js_1.TAG_ID.UNKNOWN,
- selfClosing: false,
- ackSelfClosing: false,
- attrs: [],
- location: this.getCurrentLocation(1),
- };
- }
- _createEndTagToken() {
- this.currentToken = {
- type: token_js_1.TokenType.END_TAG,
- tagName: '',
- tagID: html_js_1.TAG_ID.UNKNOWN,
- selfClosing: false,
- ackSelfClosing: false,
- attrs: [],
- location: this.getCurrentLocation(2),
- };
- }
- _createCommentToken(offset) {
- this.currentToken = {
- type: token_js_1.TokenType.COMMENT,
- data: '',
- location: this.getCurrentLocation(offset),
- };
- }
- _createDoctypeToken(initialName) {
- this.currentToken = {
- type: token_js_1.TokenType.DOCTYPE,
- name: initialName,
- forceQuirks: false,
- publicId: null,
- systemId: null,
- location: this.currentLocation,
- };
- }
- _createCharacterToken(type, chars) {
- this.currentCharacterToken = {
- type,
- chars,
- location: this.currentLocation,
- };
- }
- //Tag attributes
- _createAttr(attrNameFirstCh) {
- this.currentAttr = {
- name: attrNameFirstCh,
- value: '',
- };
- this.currentLocation = this.getCurrentLocation(0);
- }
- _leaveAttrName() {
- var _a;
- var _b;
- const token = this.currentToken;
- if ((0, token_js_1.getTokenAttr)(token, this.currentAttr.name) === null) {
- token.attrs.push(this.currentAttr);
- if (token.location && this.currentLocation) {
- const attrLocations = ((_a = (_b = token.location).attrs) !== null && _a !== void 0 ? _a : (_b.attrs = Object.create(null)));
- attrLocations[this.currentAttr.name] = this.currentLocation;
- // Set end location
- this._leaveAttrValue();
- }
- }
- else {
- this._err(error_codes_js_1.ERR.duplicateAttribute);
- }
- }
- _leaveAttrValue() {
- if (this.currentLocation) {
- this.currentLocation.endLine = this.preprocessor.line;
- this.currentLocation.endCol = this.preprocessor.col;
- this.currentLocation.endOffset = this.preprocessor.offset;
- }
- }
- //Token emission
- prepareToken(ct) {
- this._emitCurrentCharacterToken(ct.location);
- this.currentToken = null;
- if (ct.location) {
- ct.location.endLine = this.preprocessor.line;
- ct.location.endCol = this.preprocessor.col + 1;
- ct.location.endOffset = this.preprocessor.offset + 1;
- }
- this.currentLocation = this.getCurrentLocation(-1);
- }
- emitCurrentTagToken() {
- const ct = this.currentToken;
- this.prepareToken(ct);
- ct.tagID = (0, html_js_1.getTagID)(ct.tagName);
- if (ct.type === token_js_1.TokenType.START_TAG) {
- this.lastStartTagName = ct.tagName;
- this.handler.onStartTag(ct);
- }
- else {
- if (ct.attrs.length > 0) {
- this._err(error_codes_js_1.ERR.endTagWithAttributes);
- }
- if (ct.selfClosing) {
- this._err(error_codes_js_1.ERR.endTagWithTrailingSolidus);
- }
- this.handler.onEndTag(ct);
- }
- this.preprocessor.dropParsedChunk();
- }
- emitCurrentComment(ct) {
- this.prepareToken(ct);
- this.handler.onComment(ct);
- this.preprocessor.dropParsedChunk();
- }
- emitCurrentDoctype(ct) {
- this.prepareToken(ct);
- this.handler.onDoctype(ct);
- this.preprocessor.dropParsedChunk();
- }
- _emitCurrentCharacterToken(nextLocation) {
- if (this.currentCharacterToken) {
- //NOTE: if we have a pending character token, make it's end location equal to the
- //current token's start location.
- if (nextLocation && this.currentCharacterToken.location) {
- this.currentCharacterToken.location.endLine = nextLocation.startLine;
- this.currentCharacterToken.location.endCol = nextLocation.startCol;
- this.currentCharacterToken.location.endOffset = nextLocation.startOffset;
- }
- switch (this.currentCharacterToken.type) {
- case token_js_1.TokenType.CHARACTER: {
- this.handler.onCharacter(this.currentCharacterToken);
- break;
- }
- case token_js_1.TokenType.NULL_CHARACTER: {
- this.handler.onNullCharacter(this.currentCharacterToken);
- break;
- }
- case token_js_1.TokenType.WHITESPACE_CHARACTER: {
- this.handler.onWhitespaceCharacter(this.currentCharacterToken);
- break;
- }
- }
- this.currentCharacterToken = null;
- }
- }
- _emitEOFToken() {
- const location = this.getCurrentLocation(0);
- if (location) {
- location.endLine = location.startLine;
- location.endCol = location.startCol;
- location.endOffset = location.startOffset;
- }
- this._emitCurrentCharacterToken(location);
- this.handler.onEof({ type: token_js_1.TokenType.EOF, location });
- this.active = false;
- }
- //Characters emission
- //OPTIMIZATION: specification uses only one type of character tokens (one token per character).
- //This causes a huge memory overhead and a lot of unnecessary parser loops. parse5 uses 3 groups of characters.
- //If we have a sequence of characters that belong to the same group, the parser can process it
- //as a single solid character token.
- //So, there are 3 types of character tokens in parse5:
- //1)TokenType.NULL_CHARACTER - \u0000-character sequences (e.g. '\u0000\u0000\u0000')
- //2)TokenType.WHITESPACE_CHARACTER - any whitespace/new-line character sequences (e.g. '\n \r\t \f')
- //3)TokenType.CHARACTER - any character sequence which don't belong to groups 1 and 2 (e.g. 'abcdef1234@@#$%^')
- _appendCharToCurrentCharacterToken(type, ch) {
- if (this.currentCharacterToken) {
- if (this.currentCharacterToken.type !== type) {
- this.currentLocation = this.getCurrentLocation(0);
- this._emitCurrentCharacterToken(this.currentLocation);
- this.preprocessor.dropParsedChunk();
- }
- else {
- this.currentCharacterToken.chars += ch;
- return;
- }
- }
- this._createCharacterToken(type, ch);
- }
- _emitCodePoint(cp) {
- const type = isWhitespace(cp)
- ? token_js_1.TokenType.WHITESPACE_CHARACTER
- : cp === unicode_js_1.CODE_POINTS.NULL
- ? token_js_1.TokenType.NULL_CHARACTER
- : token_js_1.TokenType.CHARACTER;
- this._appendCharToCurrentCharacterToken(type, String.fromCodePoint(cp));
- }
- //NOTE: used when we emit characters explicitly.
- //This is always for non-whitespace and non-null characters, which allows us to avoid additional checks.
- _emitChars(ch) {
- this._appendCharToCurrentCharacterToken(token_js_1.TokenType.CHARACTER, ch);
- }
- // Character reference helpers
- _matchNamedCharacterReference(cp) {
- let result = null;
- let excess = 0;
- let withoutSemicolon = false;
- for (let i = 0, current = decode_js_1.htmlDecodeTree[0]; i >= 0; cp = this._consume()) {
- i = (0, decode_js_1.determineBranch)(decode_js_1.htmlDecodeTree, current, i + 1, cp);
- if (i < 0)
- break;
- excess += 1;
- current = decode_js_1.htmlDecodeTree[i];
- const masked = current & decode_js_1.BinTrieFlags.VALUE_LENGTH;
- // If the branch is a value, store it and continue
- if (masked) {
- // The mask is the number of bytes of the value, including the current byte.
- const valueLength = (masked >> 14) - 1;
- // Attribute values that aren't terminated properly aren't parsed, and shouldn't lead to a parser error.
- // See the example in https://html.spec.whatwg.org/multipage/parsing.html#named-character-reference-state
- if (cp !== unicode_js_1.CODE_POINTS.SEMICOLON &&
- this._isCharacterReferenceInAttribute() &&
- isEntityInAttributeInvalidEnd(this.preprocessor.peek(1))) {
- //NOTE: we don't flush all consumed code points here, and instead switch back to the original state after
- //emitting an ampersand. This is fine, as alphanumeric characters won't be parsed differently in attributes.
- result = [unicode_js_1.CODE_POINTS.AMPERSAND];
- // Skip over the value.
- i += valueLength;
- }
- else {
- // If this is a surrogate pair, consume the next two bytes.
- result =
- valueLength === 0
- ? [decode_js_1.htmlDecodeTree[i] & ~decode_js_1.BinTrieFlags.VALUE_LENGTH]
- : valueLength === 1
- ? [decode_js_1.htmlDecodeTree[++i]]
- : [decode_js_1.htmlDecodeTree[++i], decode_js_1.htmlDecodeTree[++i]];
- excess = 0;
- withoutSemicolon = cp !== unicode_js_1.CODE_POINTS.SEMICOLON;
- }
- if (valueLength === 0) {
- // If the value is zero-length, we're done.
- this._consume();
- break;
- }
- }
- }
- this._unconsume(excess);
- if (withoutSemicolon && !this.preprocessor.endOfChunkHit) {
- this._err(error_codes_js_1.ERR.missingSemicolonAfterCharacterReference);
- }
- // We want to emit the error above on the code point after the entity.
- // We always consume one code point too many in the loop, and we wait to
- // unconsume it until after the error is emitted.
- this._unconsume(1);
- return result;
- }
- _isCharacterReferenceInAttribute() {
- return (this.returnState === State.ATTRIBUTE_VALUE_DOUBLE_QUOTED ||
- this.returnState === State.ATTRIBUTE_VALUE_SINGLE_QUOTED ||
- this.returnState === State.ATTRIBUTE_VALUE_UNQUOTED);
- }
- _flushCodePointConsumedAsCharacterReference(cp) {
- if (this._isCharacterReferenceInAttribute()) {
- this.currentAttr.value += String.fromCodePoint(cp);
- }
- else {
- this._emitCodePoint(cp);
- }
- }
- // Calling states this way turns out to be much faster than any other approach.
- _callState(cp) {
- switch (this.state) {
- case State.DATA: {
- this._stateData(cp);
- break;
- }
- case State.RCDATA: {
- this._stateRcdata(cp);
- break;
- }
- case State.RAWTEXT: {
- this._stateRawtext(cp);
- break;
- }
- case State.SCRIPT_DATA: {
- this._stateScriptData(cp);
- break;
- }
- case State.PLAINTEXT: {
- this._statePlaintext(cp);
- break;
- }
- case State.TAG_OPEN: {
- this._stateTagOpen(cp);
- break;
- }
- case State.END_TAG_OPEN: {
- this._stateEndTagOpen(cp);
- break;
- }
- case State.TAG_NAME: {
- this._stateTagName(cp);
- break;
- }
- case State.RCDATA_LESS_THAN_SIGN: {
- this._stateRcdataLessThanSign(cp);
- break;
- }
- case State.RCDATA_END_TAG_OPEN: {
- this._stateRcdataEndTagOpen(cp);
- break;
- }
- case State.RCDATA_END_TAG_NAME: {
- this._stateRcdataEndTagName(cp);
- break;
- }
- case State.RAWTEXT_LESS_THAN_SIGN: {
- this._stateRawtextLessThanSign(cp);
- break;
- }
- case State.RAWTEXT_END_TAG_OPEN: {
- this._stateRawtextEndTagOpen(cp);
- break;
- }
- case State.RAWTEXT_END_TAG_NAME: {
- this._stateRawtextEndTagName(cp);
- break;
- }
- case State.SCRIPT_DATA_LESS_THAN_SIGN: {
- this._stateScriptDataLessThanSign(cp);
- break;
- }
- case State.SCRIPT_DATA_END_TAG_OPEN: {
- this._stateScriptDataEndTagOpen(cp);
- break;
- }
- case State.SCRIPT_DATA_END_TAG_NAME: {
- this._stateScriptDataEndTagName(cp);
- break;
- }
- case State.SCRIPT_DATA_ESCAPE_START: {
- this._stateScriptDataEscapeStart(cp);
- break;
- }
- case State.SCRIPT_DATA_ESCAPE_START_DASH: {
- this._stateScriptDataEscapeStartDash(cp);
- break;
- }
- case State.SCRIPT_DATA_ESCAPED: {
- this._stateScriptDataEscaped(cp);
- break;
- }
- case State.SCRIPT_DATA_ESCAPED_DASH: {
- this._stateScriptDataEscapedDash(cp);
- break;
- }
- case State.SCRIPT_DATA_ESCAPED_DASH_DASH: {
- this._stateScriptDataEscapedDashDash(cp);
- break;
- }
- case State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN: {
- this._stateScriptDataEscapedLessThanSign(cp);
- break;
- }
- case State.SCRIPT_DATA_ESCAPED_END_TAG_OPEN: {
- this._stateScriptDataEscapedEndTagOpen(cp);
- break;
- }
- case State.SCRIPT_DATA_ESCAPED_END_TAG_NAME: {
- this._stateScriptDataEscapedEndTagName(cp);
- break;
- }
- case State.SCRIPT_DATA_DOUBLE_ESCAPE_START: {
- this._stateScriptDataDoubleEscapeStart(cp);
- break;
- }
- case State.SCRIPT_DATA_DOUBLE_ESCAPED: {
- this._stateScriptDataDoubleEscaped(cp);
- break;
- }
- case State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH: {
- this._stateScriptDataDoubleEscapedDash(cp);
- break;
- }
- case State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH: {
- this._stateScriptDataDoubleEscapedDashDash(cp);
- break;
- }
- case State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN: {
- this._stateScriptDataDoubleEscapedLessThanSign(cp);
- break;
- }
- case State.SCRIPT_DATA_DOUBLE_ESCAPE_END: {
- this._stateScriptDataDoubleEscapeEnd(cp);
- break;
- }
- case State.BEFORE_ATTRIBUTE_NAME: {
- this._stateBeforeAttributeName(cp);
- break;
- }
- case State.ATTRIBUTE_NAME: {
- this._stateAttributeName(cp);
- break;
- }
- case State.AFTER_ATTRIBUTE_NAME: {
- this._stateAfterAttributeName(cp);
- break;
- }
- case State.BEFORE_ATTRIBUTE_VALUE: {
- this._stateBeforeAttributeValue(cp);
- break;
- }
- case State.ATTRIBUTE_VALUE_DOUBLE_QUOTED: {
- this._stateAttributeValueDoubleQuoted(cp);
- break;
- }
- case State.ATTRIBUTE_VALUE_SINGLE_QUOTED: {
- this._stateAttributeValueSingleQuoted(cp);
- break;
- }
- case State.ATTRIBUTE_VALUE_UNQUOTED: {
- this._stateAttributeValueUnquoted(cp);
- break;
- }
- case State.AFTER_ATTRIBUTE_VALUE_QUOTED: {
- this._stateAfterAttributeValueQuoted(cp);
- break;
- }
- case State.SELF_CLOSING_START_TAG: {
- this._stateSelfClosingStartTag(cp);
- break;
- }
- case State.BOGUS_COMMENT: {
- this._stateBogusComment(cp);
- break;
- }
- case State.MARKUP_DECLARATION_OPEN: {
- this._stateMarkupDeclarationOpen(cp);
- break;
- }
- case State.COMMENT_START: {
- this._stateCommentStart(cp);
- break;
- }
- case State.COMMENT_START_DASH: {
- this._stateCommentStartDash(cp);
- break;
- }
- case State.COMMENT: {
- this._stateComment(cp);
- break;
- }
- case State.COMMENT_LESS_THAN_SIGN: {
- this._stateCommentLessThanSign(cp);
- break;
- }
- case State.COMMENT_LESS_THAN_SIGN_BANG: {
- this._stateCommentLessThanSignBang(cp);
- break;
- }
- case State.COMMENT_LESS_THAN_SIGN_BANG_DASH: {
- this._stateCommentLessThanSignBangDash(cp);
- break;
- }
- case State.COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH: {
- this._stateCommentLessThanSignBangDashDash(cp);
- break;
- }
- case State.COMMENT_END_DASH: {
- this._stateCommentEndDash(cp);
- break;
- }
- case State.COMMENT_END: {
- this._stateCommentEnd(cp);
- break;
- }
- case State.COMMENT_END_BANG: {
- this._stateCommentEndBang(cp);
- break;
- }
- case State.DOCTYPE: {
- this._stateDoctype(cp);
- break;
- }
- case State.BEFORE_DOCTYPE_NAME: {
- this._stateBeforeDoctypeName(cp);
- break;
- }
- case State.DOCTYPE_NAME: {
- this._stateDoctypeName(cp);
- break;
- }
- case State.AFTER_DOCTYPE_NAME: {
- this._stateAfterDoctypeName(cp);
- break;
- }
- case State.AFTER_DOCTYPE_PUBLIC_KEYWORD: {
- this._stateAfterDoctypePublicKeyword(cp);
- break;
- }
- case State.BEFORE_DOCTYPE_PUBLIC_IDENTIFIER: {
- this._stateBeforeDoctypePublicIdentifier(cp);
- break;
- }
- case State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED: {
- this._stateDoctypePublicIdentifierDoubleQuoted(cp);
- break;
- }
- case State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED: {
- this._stateDoctypePublicIdentifierSingleQuoted(cp);
- break;
- }
- case State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER: {
- this._stateAfterDoctypePublicIdentifier(cp);
- break;
- }
- case State.BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS: {
- this._stateBetweenDoctypePublicAndSystemIdentifiers(cp);
- break;
- }
- case State.AFTER_DOCTYPE_SYSTEM_KEYWORD: {
- this._stateAfterDoctypeSystemKeyword(cp);
- break;
- }
- case State.BEFORE_DOCTYPE_SYSTEM_IDENTIFIER: {
- this._stateBeforeDoctypeSystemIdentifier(cp);
- break;
- }
- case State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED: {
- this._stateDoctypeSystemIdentifierDoubleQuoted(cp);
- break;
- }
- case State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED: {
- this._stateDoctypeSystemIdentifierSingleQuoted(cp);
- break;
- }
- case State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER: {
- this._stateAfterDoctypeSystemIdentifier(cp);
- break;
- }
- case State.BOGUS_DOCTYPE: {
- this._stateBogusDoctype(cp);
- break;
- }
- case State.CDATA_SECTION: {
- this._stateCdataSection(cp);
- break;
- }
- case State.CDATA_SECTION_BRACKET: {
- this._stateCdataSectionBracket(cp);
- break;
- }
- case State.CDATA_SECTION_END: {
- this._stateCdataSectionEnd(cp);
- break;
- }
- case State.CHARACTER_REFERENCE: {
- this._stateCharacterReference(cp);
- break;
- }
- case State.NAMED_CHARACTER_REFERENCE: {
- this._stateNamedCharacterReference(cp);
- break;
- }
- case State.AMBIGUOUS_AMPERSAND: {
- this._stateAmbiguousAmpersand(cp);
- break;
- }
- case State.NUMERIC_CHARACTER_REFERENCE: {
- this._stateNumericCharacterReference(cp);
- break;
- }
- case State.HEXADEMICAL_CHARACTER_REFERENCE_START: {
- this._stateHexademicalCharacterReferenceStart(cp);
- break;
- }
- case State.HEXADEMICAL_CHARACTER_REFERENCE: {
- this._stateHexademicalCharacterReference(cp);
- break;
- }
- case State.DECIMAL_CHARACTER_REFERENCE: {
- this._stateDecimalCharacterReference(cp);
- break;
- }
- case State.NUMERIC_CHARACTER_REFERENCE_END: {
- this._stateNumericCharacterReferenceEnd(cp);
- break;
- }
- default: {
- throw new Error('Unknown state');
- }
- }
- }
- // State machine
- // Data state
- //------------------------------------------------------------------
- _stateData(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- this.state = State.TAG_OPEN;
- break;
- }
- case unicode_js_1.CODE_POINTS.AMPERSAND: {
- this.returnState = State.DATA;
- this.state = State.CHARACTER_REFERENCE;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this._emitCodePoint(cp);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._emitEOFToken();
- break;
- }
- default: {
- this._emitCodePoint(cp);
- }
- }
- }
- // RCDATA state
- //------------------------------------------------------------------
- _stateRcdata(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.AMPERSAND: {
- this.returnState = State.RCDATA;
- this.state = State.CHARACTER_REFERENCE;
- break;
- }
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- this.state = State.RCDATA_LESS_THAN_SIGN;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._emitEOFToken();
- break;
- }
- default: {
- this._emitCodePoint(cp);
- }
- }
- }
- // RAWTEXT state
- //------------------------------------------------------------------
- _stateRawtext(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- this.state = State.RAWTEXT_LESS_THAN_SIGN;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._emitEOFToken();
- break;
- }
- default: {
- this._emitCodePoint(cp);
- }
- }
- }
- // Script data state
- //------------------------------------------------------------------
- _stateScriptData(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- this.state = State.SCRIPT_DATA_LESS_THAN_SIGN;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._emitEOFToken();
- break;
- }
- default: {
- this._emitCodePoint(cp);
- }
- }
- }
- // PLAINTEXT state
- //------------------------------------------------------------------
- _statePlaintext(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._emitEOFToken();
- break;
- }
- default: {
- this._emitCodePoint(cp);
- }
- }
- }
- // Tag open state
- //------------------------------------------------------------------
- _stateTagOpen(cp) {
- if (isAsciiLetter(cp)) {
- this._createStartTagToken();
- this.state = State.TAG_NAME;
- this._stateTagName(cp);
- }
- else
- switch (cp) {
- case unicode_js_1.CODE_POINTS.EXCLAMATION_MARK: {
- this.state = State.MARKUP_DECLARATION_OPEN;
- break;
- }
- case unicode_js_1.CODE_POINTS.SOLIDUS: {
- this.state = State.END_TAG_OPEN;
- break;
- }
- case unicode_js_1.CODE_POINTS.QUESTION_MARK: {
- this._err(error_codes_js_1.ERR.unexpectedQuestionMarkInsteadOfTagName);
- this._createCommentToken(1);
- this.state = State.BOGUS_COMMENT;
- this._stateBogusComment(cp);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofBeforeTagName);
- this._emitChars('<');
- this._emitEOFToken();
- break;
- }
- default: {
- this._err(error_codes_js_1.ERR.invalidFirstCharacterOfTagName);
- this._emitChars('<');
- this.state = State.DATA;
- this._stateData(cp);
- }
- }
- }
- // End tag open state
- //------------------------------------------------------------------
- _stateEndTagOpen(cp) {
- if (isAsciiLetter(cp)) {
- this._createEndTagToken();
- this.state = State.TAG_NAME;
- this._stateTagName(cp);
- }
- else
- switch (cp) {
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.missingEndTagName);
- this.state = State.DATA;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofBeforeTagName);
- this._emitChars('</');
- this._emitEOFToken();
- break;
- }
- default: {
- this._err(error_codes_js_1.ERR.invalidFirstCharacterOfTagName);
- this._createCommentToken(2);
- this.state = State.BOGUS_COMMENT;
- this._stateBogusComment(cp);
- }
- }
- }
- // Tag name state
- //------------------------------------------------------------------
- _stateTagName(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- this.state = State.BEFORE_ATTRIBUTE_NAME;
- break;
- }
- case unicode_js_1.CODE_POINTS.SOLIDUS: {
- this.state = State.SELF_CLOSING_START_TAG;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.state = State.DATA;
- this.emitCurrentTagToken();
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- token.tagName += unicode_js_1.REPLACEMENT_CHARACTER;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInTag);
- this._emitEOFToken();
- break;
- }
- default: {
- token.tagName += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
- }
- }
- }
- // RCDATA less-than sign state
- //------------------------------------------------------------------
- _stateRcdataLessThanSign(cp) {
- if (cp === unicode_js_1.CODE_POINTS.SOLIDUS) {
- this.state = State.RCDATA_END_TAG_OPEN;
- }
- else {
- this._emitChars('<');
- this.state = State.RCDATA;
- this._stateRcdata(cp);
- }
- }
- // RCDATA end tag open state
- //------------------------------------------------------------------
- _stateRcdataEndTagOpen(cp) {
- if (isAsciiLetter(cp)) {
- this.state = State.RCDATA_END_TAG_NAME;
- this._stateRcdataEndTagName(cp);
- }
- else {
- this._emitChars('</');
- this.state = State.RCDATA;
- this._stateRcdata(cp);
- }
- }
- handleSpecialEndTag(_cp) {
- if (!this.preprocessor.startsWith(this.lastStartTagName, false)) {
- return !this._ensureHibernation();
- }
- this._createEndTagToken();
- const token = this.currentToken;
- token.tagName = this.lastStartTagName;
- const cp = this.preprocessor.peek(this.lastStartTagName.length);
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- this._advanceBy(this.lastStartTagName.length);
- this.state = State.BEFORE_ATTRIBUTE_NAME;
- return false;
- }
- case unicode_js_1.CODE_POINTS.SOLIDUS: {
- this._advanceBy(this.lastStartTagName.length);
- this.state = State.SELF_CLOSING_START_TAG;
- return false;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._advanceBy(this.lastStartTagName.length);
- this.emitCurrentTagToken();
- this.state = State.DATA;
- return false;
- }
- default: {
- return !this._ensureHibernation();
- }
- }
- }
- // RCDATA end tag name state
- //------------------------------------------------------------------
- _stateRcdataEndTagName(cp) {
- if (this.handleSpecialEndTag(cp)) {
- this._emitChars('</');
- this.state = State.RCDATA;
- this._stateRcdata(cp);
- }
- }
- // RAWTEXT less-than sign state
- //------------------------------------------------------------------
- _stateRawtextLessThanSign(cp) {
- if (cp === unicode_js_1.CODE_POINTS.SOLIDUS) {
- this.state = State.RAWTEXT_END_TAG_OPEN;
- }
- else {
- this._emitChars('<');
- this.state = State.RAWTEXT;
- this._stateRawtext(cp);
- }
- }
- // RAWTEXT end tag open state
- //------------------------------------------------------------------
- _stateRawtextEndTagOpen(cp) {
- if (isAsciiLetter(cp)) {
- this.state = State.RAWTEXT_END_TAG_NAME;
- this._stateRawtextEndTagName(cp);
- }
- else {
- this._emitChars('</');
- this.state = State.RAWTEXT;
- this._stateRawtext(cp);
- }
- }
- // RAWTEXT end tag name state
- //------------------------------------------------------------------
- _stateRawtextEndTagName(cp) {
- if (this.handleSpecialEndTag(cp)) {
- this._emitChars('</');
- this.state = State.RAWTEXT;
- this._stateRawtext(cp);
- }
- }
- // Script data less-than sign state
- //------------------------------------------------------------------
- _stateScriptDataLessThanSign(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SOLIDUS: {
- this.state = State.SCRIPT_DATA_END_TAG_OPEN;
- break;
- }
- case unicode_js_1.CODE_POINTS.EXCLAMATION_MARK: {
- this.state = State.SCRIPT_DATA_ESCAPE_START;
- this._emitChars('<!');
- break;
- }
- default: {
- this._emitChars('<');
- this.state = State.SCRIPT_DATA;
- this._stateScriptData(cp);
- }
- }
- }
- // Script data end tag open state
- //------------------------------------------------------------------
- _stateScriptDataEndTagOpen(cp) {
- if (isAsciiLetter(cp)) {
- this.state = State.SCRIPT_DATA_END_TAG_NAME;
- this._stateScriptDataEndTagName(cp);
- }
- else {
- this._emitChars('</');
- this.state = State.SCRIPT_DATA;
- this._stateScriptData(cp);
- }
- }
- // Script data end tag name state
- //------------------------------------------------------------------
- _stateScriptDataEndTagName(cp) {
- if (this.handleSpecialEndTag(cp)) {
- this._emitChars('</');
- this.state = State.SCRIPT_DATA;
- this._stateScriptData(cp);
- }
- }
- // Script data escape start state
- //------------------------------------------------------------------
- _stateScriptDataEscapeStart(cp) {
- if (cp === unicode_js_1.CODE_POINTS.HYPHEN_MINUS) {
- this.state = State.SCRIPT_DATA_ESCAPE_START_DASH;
- this._emitChars('-');
- }
- else {
- this.state = State.SCRIPT_DATA;
- this._stateScriptData(cp);
- }
- }
- // Script data escape start dash state
- //------------------------------------------------------------------
- _stateScriptDataEscapeStartDash(cp) {
- if (cp === unicode_js_1.CODE_POINTS.HYPHEN_MINUS) {
- this.state = State.SCRIPT_DATA_ESCAPED_DASH_DASH;
- this._emitChars('-');
- }
- else {
- this.state = State.SCRIPT_DATA;
- this._stateScriptData(cp);
- }
- }
- // Script data escaped state
- //------------------------------------------------------------------
- _stateScriptDataEscaped(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
- this.state = State.SCRIPT_DATA_ESCAPED_DASH;
- this._emitChars('-');
- break;
- }
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInScriptHtmlCommentLikeText);
- this._emitEOFToken();
- break;
- }
- default: {
- this._emitCodePoint(cp);
- }
- }
- }
- // Script data escaped dash state
- //------------------------------------------------------------------
- _stateScriptDataEscapedDash(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
- this.state = State.SCRIPT_DATA_ESCAPED_DASH_DASH;
- this._emitChars('-');
- break;
- }
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this.state = State.SCRIPT_DATA_ESCAPED;
- this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInScriptHtmlCommentLikeText);
- this._emitEOFToken();
- break;
- }
- default: {
- this.state = State.SCRIPT_DATA_ESCAPED;
- this._emitCodePoint(cp);
- }
- }
- }
- // Script data escaped dash dash state
- //------------------------------------------------------------------
- _stateScriptDataEscapedDashDash(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
- this._emitChars('-');
- break;
- }
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.state = State.SCRIPT_DATA;
- this._emitChars('>');
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this.state = State.SCRIPT_DATA_ESCAPED;
- this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInScriptHtmlCommentLikeText);
- this._emitEOFToken();
- break;
- }
- default: {
- this.state = State.SCRIPT_DATA_ESCAPED;
- this._emitCodePoint(cp);
- }
- }
- }
- // Script data escaped less-than sign state
- //------------------------------------------------------------------
- _stateScriptDataEscapedLessThanSign(cp) {
- if (cp === unicode_js_1.CODE_POINTS.SOLIDUS) {
- this.state = State.SCRIPT_DATA_ESCAPED_END_TAG_OPEN;
- }
- else if (isAsciiLetter(cp)) {
- this._emitChars('<');
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPE_START;
- this._stateScriptDataDoubleEscapeStart(cp);
- }
- else {
- this._emitChars('<');
- this.state = State.SCRIPT_DATA_ESCAPED;
- this._stateScriptDataEscaped(cp);
- }
- }
- // Script data escaped end tag open state
- //------------------------------------------------------------------
- _stateScriptDataEscapedEndTagOpen(cp) {
- if (isAsciiLetter(cp)) {
- this.state = State.SCRIPT_DATA_ESCAPED_END_TAG_NAME;
- this._stateScriptDataEscapedEndTagName(cp);
- }
- else {
- this._emitChars('</');
- this.state = State.SCRIPT_DATA_ESCAPED;
- this._stateScriptDataEscaped(cp);
- }
- }
- // Script data escaped end tag name state
- //------------------------------------------------------------------
- _stateScriptDataEscapedEndTagName(cp) {
- if (this.handleSpecialEndTag(cp)) {
- this._emitChars('</');
- this.state = State.SCRIPT_DATA_ESCAPED;
- this._stateScriptDataEscaped(cp);
- }
- }
- // Script data double escape start state
- //------------------------------------------------------------------
- _stateScriptDataDoubleEscapeStart(cp) {
- if (this.preprocessor.startsWith(unicode_js_1.SEQUENCES.SCRIPT, false) &&
- isScriptDataDoubleEscapeSequenceEnd(this.preprocessor.peek(unicode_js_1.SEQUENCES.SCRIPT.length))) {
- this._emitCodePoint(cp);
- for (let i = 0; i < unicode_js_1.SEQUENCES.SCRIPT.length; i++) {
- this._emitCodePoint(this._consume());
- }
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
- }
- else if (!this._ensureHibernation()) {
- this.state = State.SCRIPT_DATA_ESCAPED;
- this._stateScriptDataEscaped(cp);
- }
- }
- // Script data double escaped state
- //------------------------------------------------------------------
- _stateScriptDataDoubleEscaped(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH;
- this._emitChars('-');
- break;
- }
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
- this._emitChars('<');
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInScriptHtmlCommentLikeText);
- this._emitEOFToken();
- break;
- }
- default: {
- this._emitCodePoint(cp);
- }
- }
- }
- // Script data double escaped dash state
- //------------------------------------------------------------------
- _stateScriptDataDoubleEscapedDash(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH;
- this._emitChars('-');
- break;
- }
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
- this._emitChars('<');
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
- this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInScriptHtmlCommentLikeText);
- this._emitEOFToken();
- break;
- }
- default: {
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
- this._emitCodePoint(cp);
- }
- }
- }
- // Script data double escaped dash dash state
- //------------------------------------------------------------------
- _stateScriptDataDoubleEscapedDashDash(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
- this._emitChars('-');
- break;
- }
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
- this._emitChars('<');
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.state = State.SCRIPT_DATA;
- this._emitChars('>');
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
- this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInScriptHtmlCommentLikeText);
- this._emitEOFToken();
- break;
- }
- default: {
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
- this._emitCodePoint(cp);
- }
- }
- }
- // Script data double escaped less-than sign state
- //------------------------------------------------------------------
- _stateScriptDataDoubleEscapedLessThanSign(cp) {
- if (cp === unicode_js_1.CODE_POINTS.SOLIDUS) {
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPE_END;
- this._emitChars('/');
- }
- else {
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
- this._stateScriptDataDoubleEscaped(cp);
- }
- }
- // Script data double escape end state
- //------------------------------------------------------------------
- _stateScriptDataDoubleEscapeEnd(cp) {
- if (this.preprocessor.startsWith(unicode_js_1.SEQUENCES.SCRIPT, false) &&
- isScriptDataDoubleEscapeSequenceEnd(this.preprocessor.peek(unicode_js_1.SEQUENCES.SCRIPT.length))) {
- this._emitCodePoint(cp);
- for (let i = 0; i < unicode_js_1.SEQUENCES.SCRIPT.length; i++) {
- this._emitCodePoint(this._consume());
- }
- this.state = State.SCRIPT_DATA_ESCAPED;
- }
- else if (!this._ensureHibernation()) {
- this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
- this._stateScriptDataDoubleEscaped(cp);
- }
- }
- // Before attribute name state
- //------------------------------------------------------------------
- _stateBeforeAttributeName(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- // Ignore whitespace
- break;
- }
- case unicode_js_1.CODE_POINTS.SOLIDUS:
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN:
- case unicode_js_1.CODE_POINTS.EOF: {
- this.state = State.AFTER_ATTRIBUTE_NAME;
- this._stateAfterAttributeName(cp);
- break;
- }
- case unicode_js_1.CODE_POINTS.EQUALS_SIGN: {
- this._err(error_codes_js_1.ERR.unexpectedEqualsSignBeforeAttributeName);
- this._createAttr('=');
- this.state = State.ATTRIBUTE_NAME;
- break;
- }
- default: {
- this._createAttr('');
- this.state = State.ATTRIBUTE_NAME;
- this._stateAttributeName(cp);
- }
- }
- }
- // Attribute name state
- //------------------------------------------------------------------
- _stateAttributeName(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED:
- case unicode_js_1.CODE_POINTS.SOLIDUS:
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN:
- case unicode_js_1.CODE_POINTS.EOF: {
- this._leaveAttrName();
- this.state = State.AFTER_ATTRIBUTE_NAME;
- this._stateAfterAttributeName(cp);
- break;
- }
- case unicode_js_1.CODE_POINTS.EQUALS_SIGN: {
- this._leaveAttrName();
- this.state = State.BEFORE_ATTRIBUTE_VALUE;
- break;
- }
- case unicode_js_1.CODE_POINTS.QUOTATION_MARK:
- case unicode_js_1.CODE_POINTS.APOSTROPHE:
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.unexpectedCharacterInAttributeName);
- this.currentAttr.name += String.fromCodePoint(cp);
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this.currentAttr.name += unicode_js_1.REPLACEMENT_CHARACTER;
- break;
- }
- default: {
- this.currentAttr.name += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
- }
- }
- }
- // After attribute name state
- //------------------------------------------------------------------
- _stateAfterAttributeName(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- // Ignore whitespace
- break;
- }
- case unicode_js_1.CODE_POINTS.SOLIDUS: {
- this.state = State.SELF_CLOSING_START_TAG;
- break;
- }
- case unicode_js_1.CODE_POINTS.EQUALS_SIGN: {
- this.state = State.BEFORE_ATTRIBUTE_VALUE;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.state = State.DATA;
- this.emitCurrentTagToken();
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInTag);
- this._emitEOFToken();
- break;
- }
- default: {
- this._createAttr('');
- this.state = State.ATTRIBUTE_NAME;
- this._stateAttributeName(cp);
- }
- }
- }
- // Before attribute value state
- //------------------------------------------------------------------
- _stateBeforeAttributeValue(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- // Ignore whitespace
- break;
- }
- case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
- this.state = State.ATTRIBUTE_VALUE_DOUBLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.APOSTROPHE: {
- this.state = State.ATTRIBUTE_VALUE_SINGLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.missingAttributeValue);
- this.state = State.DATA;
- this.emitCurrentTagToken();
- break;
- }
- default: {
- this.state = State.ATTRIBUTE_VALUE_UNQUOTED;
- this._stateAttributeValueUnquoted(cp);
- }
- }
- }
- // Attribute value (double-quoted) state
- //------------------------------------------------------------------
- _stateAttributeValueDoubleQuoted(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
- this.state = State.AFTER_ATTRIBUTE_VALUE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.AMPERSAND: {
- this.returnState = State.ATTRIBUTE_VALUE_DOUBLE_QUOTED;
- this.state = State.CHARACTER_REFERENCE;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this.currentAttr.value += unicode_js_1.REPLACEMENT_CHARACTER;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInTag);
- this._emitEOFToken();
- break;
- }
- default: {
- this.currentAttr.value += String.fromCodePoint(cp);
- }
- }
- }
- // Attribute value (single-quoted) state
- //------------------------------------------------------------------
- _stateAttributeValueSingleQuoted(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.APOSTROPHE: {
- this.state = State.AFTER_ATTRIBUTE_VALUE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.AMPERSAND: {
- this.returnState = State.ATTRIBUTE_VALUE_SINGLE_QUOTED;
- this.state = State.CHARACTER_REFERENCE;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this.currentAttr.value += unicode_js_1.REPLACEMENT_CHARACTER;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInTag);
- this._emitEOFToken();
- break;
- }
- default: {
- this.currentAttr.value += String.fromCodePoint(cp);
- }
- }
- }
- // Attribute value (unquoted) state
- //------------------------------------------------------------------
- _stateAttributeValueUnquoted(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- this._leaveAttrValue();
- this.state = State.BEFORE_ATTRIBUTE_NAME;
- break;
- }
- case unicode_js_1.CODE_POINTS.AMPERSAND: {
- this.returnState = State.ATTRIBUTE_VALUE_UNQUOTED;
- this.state = State.CHARACTER_REFERENCE;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._leaveAttrValue();
- this.state = State.DATA;
- this.emitCurrentTagToken();
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this.currentAttr.value += unicode_js_1.REPLACEMENT_CHARACTER;
- break;
- }
- case unicode_js_1.CODE_POINTS.QUOTATION_MARK:
- case unicode_js_1.CODE_POINTS.APOSTROPHE:
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN:
- case unicode_js_1.CODE_POINTS.EQUALS_SIGN:
- case unicode_js_1.CODE_POINTS.GRAVE_ACCENT: {
- this._err(error_codes_js_1.ERR.unexpectedCharacterInUnquotedAttributeValue);
- this.currentAttr.value += String.fromCodePoint(cp);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInTag);
- this._emitEOFToken();
- break;
- }
- default: {
- this.currentAttr.value += String.fromCodePoint(cp);
- }
- }
- }
- // After attribute value (quoted) state
- //------------------------------------------------------------------
- _stateAfterAttributeValueQuoted(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- this._leaveAttrValue();
- this.state = State.BEFORE_ATTRIBUTE_NAME;
- break;
- }
- case unicode_js_1.CODE_POINTS.SOLIDUS: {
- this._leaveAttrValue();
- this.state = State.SELF_CLOSING_START_TAG;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._leaveAttrValue();
- this.state = State.DATA;
- this.emitCurrentTagToken();
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInTag);
- this._emitEOFToken();
- break;
- }
- default: {
- this._err(error_codes_js_1.ERR.missingWhitespaceBetweenAttributes);
- this.state = State.BEFORE_ATTRIBUTE_NAME;
- this._stateBeforeAttributeName(cp);
- }
- }
- }
- // Self-closing start tag state
- //------------------------------------------------------------------
- _stateSelfClosingStartTag(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- const token = this.currentToken;
- token.selfClosing = true;
- this.state = State.DATA;
- this.emitCurrentTagToken();
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInTag);
- this._emitEOFToken();
- break;
- }
- default: {
- this._err(error_codes_js_1.ERR.unexpectedSolidusInTag);
- this.state = State.BEFORE_ATTRIBUTE_NAME;
- this._stateBeforeAttributeName(cp);
- }
- }
- }
- // Bogus comment state
- //------------------------------------------------------------------
- _stateBogusComment(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.state = State.DATA;
- this.emitCurrentComment(token);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this.emitCurrentComment(token);
- this._emitEOFToken();
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- token.data += unicode_js_1.REPLACEMENT_CHARACTER;
- break;
- }
- default: {
- token.data += String.fromCodePoint(cp);
- }
- }
- }
- // Markup declaration open state
- //------------------------------------------------------------------
- _stateMarkupDeclarationOpen(cp) {
- if (this._consumeSequenceIfMatch(unicode_js_1.SEQUENCES.DASH_DASH, true)) {
- this._createCommentToken(unicode_js_1.SEQUENCES.DASH_DASH.length + 1);
- this.state = State.COMMENT_START;
- }
- else if (this._consumeSequenceIfMatch(unicode_js_1.SEQUENCES.DOCTYPE, false)) {
- // NOTE: Doctypes tokens are created without fixed offsets. We keep track of the moment a doctype *might* start here.
- this.currentLocation = this.getCurrentLocation(unicode_js_1.SEQUENCES.DOCTYPE.length + 1);
- this.state = State.DOCTYPE;
- }
- else if (this._consumeSequenceIfMatch(unicode_js_1.SEQUENCES.CDATA_START, true)) {
- if (this.inForeignNode) {
- this.state = State.CDATA_SECTION;
- }
- else {
- this._err(error_codes_js_1.ERR.cdataInHtmlContent);
- this._createCommentToken(unicode_js_1.SEQUENCES.CDATA_START.length + 1);
- this.currentToken.data = '[CDATA[';
- this.state = State.BOGUS_COMMENT;
- }
- }
- //NOTE: Sequence lookups can be abrupted by hibernation. In that case, lookup
- //results are no longer valid and we will need to start over.
- else if (!this._ensureHibernation()) {
- this._err(error_codes_js_1.ERR.incorrectlyOpenedComment);
- this._createCommentToken(2);
- this.state = State.BOGUS_COMMENT;
- this._stateBogusComment(cp);
- }
- }
- // Comment start state
- //------------------------------------------------------------------
- _stateCommentStart(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
- this.state = State.COMMENT_START_DASH;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.abruptClosingOfEmptyComment);
- this.state = State.DATA;
- const token = this.currentToken;
- this.emitCurrentComment(token);
- break;
- }
- default: {
- this.state = State.COMMENT;
- this._stateComment(cp);
- }
- }
- }
- // Comment start dash state
- //------------------------------------------------------------------
- _stateCommentStartDash(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
- this.state = State.COMMENT_END;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.abruptClosingOfEmptyComment);
- this.state = State.DATA;
- this.emitCurrentComment(token);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInComment);
- this.emitCurrentComment(token);
- this._emitEOFToken();
- break;
- }
- default: {
- token.data += '-';
- this.state = State.COMMENT;
- this._stateComment(cp);
- }
- }
- }
- // Comment state
- //------------------------------------------------------------------
- _stateComment(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
- this.state = State.COMMENT_END_DASH;
- break;
- }
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- token.data += '<';
- this.state = State.COMMENT_LESS_THAN_SIGN;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- token.data += unicode_js_1.REPLACEMENT_CHARACTER;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInComment);
- this.emitCurrentComment(token);
- this._emitEOFToken();
- break;
- }
- default: {
- token.data += String.fromCodePoint(cp);
- }
- }
- }
- // Comment less-than sign state
- //------------------------------------------------------------------
- _stateCommentLessThanSign(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.EXCLAMATION_MARK: {
- token.data += '!';
- this.state = State.COMMENT_LESS_THAN_SIGN_BANG;
- break;
- }
- case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
- token.data += '<';
- break;
- }
- default: {
- this.state = State.COMMENT;
- this._stateComment(cp);
- }
- }
- }
- // Comment less-than sign bang state
- //------------------------------------------------------------------
- _stateCommentLessThanSignBang(cp) {
- if (cp === unicode_js_1.CODE_POINTS.HYPHEN_MINUS) {
- this.state = State.COMMENT_LESS_THAN_SIGN_BANG_DASH;
- }
- else {
- this.state = State.COMMENT;
- this._stateComment(cp);
- }
- }
- // Comment less-than sign bang dash state
- //------------------------------------------------------------------
- _stateCommentLessThanSignBangDash(cp) {
- if (cp === unicode_js_1.CODE_POINTS.HYPHEN_MINUS) {
- this.state = State.COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH;
- }
- else {
- this.state = State.COMMENT_END_DASH;
- this._stateCommentEndDash(cp);
- }
- }
- // Comment less-than sign bang dash dash state
- //------------------------------------------------------------------
- _stateCommentLessThanSignBangDashDash(cp) {
- if (cp !== unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN && cp !== unicode_js_1.CODE_POINTS.EOF) {
- this._err(error_codes_js_1.ERR.nestedComment);
- }
- this.state = State.COMMENT_END;
- this._stateCommentEnd(cp);
- }
- // Comment end dash state
- //------------------------------------------------------------------
- _stateCommentEndDash(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
- this.state = State.COMMENT_END;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInComment);
- this.emitCurrentComment(token);
- this._emitEOFToken();
- break;
- }
- default: {
- token.data += '-';
- this.state = State.COMMENT;
- this._stateComment(cp);
- }
- }
- }
- // Comment end state
- //------------------------------------------------------------------
- _stateCommentEnd(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.state = State.DATA;
- this.emitCurrentComment(token);
- break;
- }
- case unicode_js_1.CODE_POINTS.EXCLAMATION_MARK: {
- this.state = State.COMMENT_END_BANG;
- break;
- }
- case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
- token.data += '-';
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInComment);
- this.emitCurrentComment(token);
- this._emitEOFToken();
- break;
- }
- default: {
- token.data += '--';
- this.state = State.COMMENT;
- this._stateComment(cp);
- }
- }
- }
- // Comment end bang state
- //------------------------------------------------------------------
- _stateCommentEndBang(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
- token.data += '--!';
- this.state = State.COMMENT_END_DASH;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.incorrectlyClosedComment);
- this.state = State.DATA;
- this.emitCurrentComment(token);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInComment);
- this.emitCurrentComment(token);
- this._emitEOFToken();
- break;
- }
- default: {
- token.data += '--!';
- this.state = State.COMMENT;
- this._stateComment(cp);
- }
- }
- }
- // DOCTYPE state
- //------------------------------------------------------------------
- _stateDoctype(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- this.state = State.BEFORE_DOCTYPE_NAME;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.state = State.BEFORE_DOCTYPE_NAME;
- this._stateBeforeDoctypeName(cp);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- this._createDoctypeToken(null);
- const token = this.currentToken;
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- this._err(error_codes_js_1.ERR.missingWhitespaceBeforeDoctypeName);
- this.state = State.BEFORE_DOCTYPE_NAME;
- this._stateBeforeDoctypeName(cp);
- }
- }
- }
- // Before DOCTYPE name state
- //------------------------------------------------------------------
- _stateBeforeDoctypeName(cp) {
- if (isAsciiUpper(cp)) {
- this._createDoctypeToken(String.fromCharCode(toAsciiLower(cp)));
- this.state = State.DOCTYPE_NAME;
- }
- else
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- // Ignore whitespace
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- this._createDoctypeToken(unicode_js_1.REPLACEMENT_CHARACTER);
- this.state = State.DOCTYPE_NAME;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.missingDoctypeName);
- this._createDoctypeToken(null);
- const token = this.currentToken;
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this.state = State.DATA;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- this._createDoctypeToken(null);
- const token = this.currentToken;
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- this._createDoctypeToken(String.fromCodePoint(cp));
- this.state = State.DOCTYPE_NAME;
- }
- }
- }
- // DOCTYPE name state
- //------------------------------------------------------------------
- _stateDoctypeName(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- this.state = State.AFTER_DOCTYPE_NAME;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.state = State.DATA;
- this.emitCurrentDoctype(token);
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- token.name += unicode_js_1.REPLACEMENT_CHARACTER;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- token.name += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
- }
- }
- }
- // After DOCTYPE name state
- //------------------------------------------------------------------
- _stateAfterDoctypeName(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- // Ignore whitespace
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.state = State.DATA;
- this.emitCurrentDoctype(token);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- if (this._consumeSequenceIfMatch(unicode_js_1.SEQUENCES.PUBLIC, false)) {
- this.state = State.AFTER_DOCTYPE_PUBLIC_KEYWORD;
- }
- else if (this._consumeSequenceIfMatch(unicode_js_1.SEQUENCES.SYSTEM, false)) {
- this.state = State.AFTER_DOCTYPE_SYSTEM_KEYWORD;
- }
- //NOTE: sequence lookup can be abrupted by hibernation. In that case lookup
- //results are no longer valid and we will need to start over.
- else if (!this._ensureHibernation()) {
- this._err(error_codes_js_1.ERR.invalidCharacterSequenceAfterDoctypeName);
- token.forceQuirks = true;
- this.state = State.BOGUS_DOCTYPE;
- this._stateBogusDoctype(cp);
- }
- }
- }
- }
- // After DOCTYPE public keyword state
- //------------------------------------------------------------------
- _stateAfterDoctypePublicKeyword(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- this.state = State.BEFORE_DOCTYPE_PUBLIC_IDENTIFIER;
- break;
- }
- case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
- this._err(error_codes_js_1.ERR.missingWhitespaceAfterDoctypePublicKeyword);
- token.publicId = '';
- this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.APOSTROPHE: {
- this._err(error_codes_js_1.ERR.missingWhitespaceAfterDoctypePublicKeyword);
- token.publicId = '';
- this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.missingDoctypePublicIdentifier);
- token.forceQuirks = true;
- this.state = State.DATA;
- this.emitCurrentDoctype(token);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- this._err(error_codes_js_1.ERR.missingQuoteBeforeDoctypePublicIdentifier);
- token.forceQuirks = true;
- this.state = State.BOGUS_DOCTYPE;
- this._stateBogusDoctype(cp);
- }
- }
- }
- // Before DOCTYPE public identifier state
- //------------------------------------------------------------------
- _stateBeforeDoctypePublicIdentifier(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- // Ignore whitespace
- break;
- }
- case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
- token.publicId = '';
- this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.APOSTROPHE: {
- token.publicId = '';
- this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.missingDoctypePublicIdentifier);
- token.forceQuirks = true;
- this.state = State.DATA;
- this.emitCurrentDoctype(token);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- this._err(error_codes_js_1.ERR.missingQuoteBeforeDoctypePublicIdentifier);
- token.forceQuirks = true;
- this.state = State.BOGUS_DOCTYPE;
- this._stateBogusDoctype(cp);
- }
- }
- }
- // DOCTYPE public identifier (double-quoted) state
- //------------------------------------------------------------------
- _stateDoctypePublicIdentifierDoubleQuoted(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
- this.state = State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- token.publicId += unicode_js_1.REPLACEMENT_CHARACTER;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.abruptDoctypePublicIdentifier);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this.state = State.DATA;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- token.publicId += String.fromCodePoint(cp);
- }
- }
- }
- // DOCTYPE public identifier (single-quoted) state
- //------------------------------------------------------------------
- _stateDoctypePublicIdentifierSingleQuoted(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.APOSTROPHE: {
- this.state = State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- token.publicId += unicode_js_1.REPLACEMENT_CHARACTER;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.abruptDoctypePublicIdentifier);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this.state = State.DATA;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- token.publicId += String.fromCodePoint(cp);
- }
- }
- }
- // After DOCTYPE public identifier state
- //------------------------------------------------------------------
- _stateAfterDoctypePublicIdentifier(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- this.state = State.BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.state = State.DATA;
- this.emitCurrentDoctype(token);
- break;
- }
- case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
- this._err(error_codes_js_1.ERR.missingWhitespaceBetweenDoctypePublicAndSystemIdentifiers);
- token.systemId = '';
- this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.APOSTROPHE: {
- this._err(error_codes_js_1.ERR.missingWhitespaceBetweenDoctypePublicAndSystemIdentifiers);
- token.systemId = '';
- this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- this._err(error_codes_js_1.ERR.missingQuoteBeforeDoctypeSystemIdentifier);
- token.forceQuirks = true;
- this.state = State.BOGUS_DOCTYPE;
- this._stateBogusDoctype(cp);
- }
- }
- }
- // Between DOCTYPE public and system identifiers state
- //------------------------------------------------------------------
- _stateBetweenDoctypePublicAndSystemIdentifiers(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- // Ignore whitespace
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.emitCurrentDoctype(token);
- this.state = State.DATA;
- break;
- }
- case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
- token.systemId = '';
- this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.APOSTROPHE: {
- token.systemId = '';
- this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- this._err(error_codes_js_1.ERR.missingQuoteBeforeDoctypeSystemIdentifier);
- token.forceQuirks = true;
- this.state = State.BOGUS_DOCTYPE;
- this._stateBogusDoctype(cp);
- }
- }
- }
- // After DOCTYPE system keyword state
- //------------------------------------------------------------------
- _stateAfterDoctypeSystemKeyword(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- this.state = State.BEFORE_DOCTYPE_SYSTEM_IDENTIFIER;
- break;
- }
- case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
- this._err(error_codes_js_1.ERR.missingWhitespaceAfterDoctypeSystemKeyword);
- token.systemId = '';
- this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.APOSTROPHE: {
- this._err(error_codes_js_1.ERR.missingWhitespaceAfterDoctypeSystemKeyword);
- token.systemId = '';
- this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.missingDoctypeSystemIdentifier);
- token.forceQuirks = true;
- this.state = State.DATA;
- this.emitCurrentDoctype(token);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- this._err(error_codes_js_1.ERR.missingQuoteBeforeDoctypeSystemIdentifier);
- token.forceQuirks = true;
- this.state = State.BOGUS_DOCTYPE;
- this._stateBogusDoctype(cp);
- }
- }
- }
- // Before DOCTYPE system identifier state
- //------------------------------------------------------------------
- _stateBeforeDoctypeSystemIdentifier(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- // Ignore whitespace
- break;
- }
- case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
- token.systemId = '';
- this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.APOSTROPHE: {
- token.systemId = '';
- this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.missingDoctypeSystemIdentifier);
- token.forceQuirks = true;
- this.state = State.DATA;
- this.emitCurrentDoctype(token);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- this._err(error_codes_js_1.ERR.missingQuoteBeforeDoctypeSystemIdentifier);
- token.forceQuirks = true;
- this.state = State.BOGUS_DOCTYPE;
- this._stateBogusDoctype(cp);
- }
- }
- }
- // DOCTYPE system identifier (double-quoted) state
- //------------------------------------------------------------------
- _stateDoctypeSystemIdentifierDoubleQuoted(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
- this.state = State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- token.systemId += unicode_js_1.REPLACEMENT_CHARACTER;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.abruptDoctypeSystemIdentifier);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this.state = State.DATA;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- token.systemId += String.fromCodePoint(cp);
- }
- }
- }
- // DOCTYPE system identifier (single-quoted) state
- //------------------------------------------------------------------
- _stateDoctypeSystemIdentifierSingleQuoted(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.APOSTROPHE: {
- this.state = State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- token.systemId += unicode_js_1.REPLACEMENT_CHARACTER;
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this._err(error_codes_js_1.ERR.abruptDoctypeSystemIdentifier);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this.state = State.DATA;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- token.systemId += String.fromCodePoint(cp);
- }
- }
- }
- // After DOCTYPE system identifier state
- //------------------------------------------------------------------
- _stateAfterDoctypeSystemIdentifier(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.SPACE:
- case unicode_js_1.CODE_POINTS.LINE_FEED:
- case unicode_js_1.CODE_POINTS.TABULATION:
- case unicode_js_1.CODE_POINTS.FORM_FEED: {
- // Ignore whitespace
- break;
- }
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.emitCurrentDoctype(token);
- this.state = State.DATA;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInDoctype);
- token.forceQuirks = true;
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default: {
- this._err(error_codes_js_1.ERR.unexpectedCharacterAfterDoctypeSystemIdentifier);
- this.state = State.BOGUS_DOCTYPE;
- this._stateBogusDoctype(cp);
- }
- }
- }
- // Bogus DOCTYPE state
- //------------------------------------------------------------------
- _stateBogusDoctype(cp) {
- const token = this.currentToken;
- switch (cp) {
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.emitCurrentDoctype(token);
- this.state = State.DATA;
- break;
- }
- case unicode_js_1.CODE_POINTS.NULL: {
- this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this.emitCurrentDoctype(token);
- this._emitEOFToken();
- break;
- }
- default:
- // Do nothing
- }
- }
- // CDATA section state
- //------------------------------------------------------------------
- _stateCdataSection(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.RIGHT_SQUARE_BRACKET: {
- this.state = State.CDATA_SECTION_BRACKET;
- break;
- }
- case unicode_js_1.CODE_POINTS.EOF: {
- this._err(error_codes_js_1.ERR.eofInCdata);
- this._emitEOFToken();
- break;
- }
- default: {
- this._emitCodePoint(cp);
- }
- }
- }
- // CDATA section bracket state
- //------------------------------------------------------------------
- _stateCdataSectionBracket(cp) {
- if (cp === unicode_js_1.CODE_POINTS.RIGHT_SQUARE_BRACKET) {
- this.state = State.CDATA_SECTION_END;
- }
- else {
- this._emitChars(']');
- this.state = State.CDATA_SECTION;
- this._stateCdataSection(cp);
- }
- }
- // CDATA section end state
- //------------------------------------------------------------------
- _stateCdataSectionEnd(cp) {
- switch (cp) {
- case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
- this.state = State.DATA;
- break;
- }
- case unicode_js_1.CODE_POINTS.RIGHT_SQUARE_BRACKET: {
- this._emitChars(']');
- break;
- }
- default: {
- this._emitChars(']]');
- this.state = State.CDATA_SECTION;
- this._stateCdataSection(cp);
- }
- }
- }
- // Character reference state
- //------------------------------------------------------------------
- _stateCharacterReference(cp) {
- if (cp === unicode_js_1.CODE_POINTS.NUMBER_SIGN) {
- this.state = State.NUMERIC_CHARACTER_REFERENCE;
- }
- else if (isAsciiAlphaNumeric(cp)) {
- this.state = State.NAMED_CHARACTER_REFERENCE;
- this._stateNamedCharacterReference(cp);
- }
- else {
- this._flushCodePointConsumedAsCharacterReference(unicode_js_1.CODE_POINTS.AMPERSAND);
- this._reconsumeInState(this.returnState, cp);
- }
- }
- // Named character reference state
- //------------------------------------------------------------------
- _stateNamedCharacterReference(cp) {
- const matchResult = this._matchNamedCharacterReference(cp);
- //NOTE: Matching can be abrupted by hibernation. In that case, match
- //results are no longer valid and we will need to start over.
- if (this._ensureHibernation()) {
- // Stay in the state, try again.
- }
- else if (matchResult) {
- for (let i = 0; i < matchResult.length; i++) {
- this._flushCodePointConsumedAsCharacterReference(matchResult[i]);
- }
- this.state = this.returnState;
- }
- else {
- this._flushCodePointConsumedAsCharacterReference(unicode_js_1.CODE_POINTS.AMPERSAND);
- this.state = State.AMBIGUOUS_AMPERSAND;
- }
- }
- // Ambiguos ampersand state
- //------------------------------------------------------------------
- _stateAmbiguousAmpersand(cp) {
- if (isAsciiAlphaNumeric(cp)) {
- this._flushCodePointConsumedAsCharacterReference(cp);
- }
- else {
- if (cp === unicode_js_1.CODE_POINTS.SEMICOLON) {
- this._err(error_codes_js_1.ERR.unknownNamedCharacterReference);
- }
- this._reconsumeInState(this.returnState, cp);
- }
- }
- // Numeric character reference state
- //------------------------------------------------------------------
- _stateNumericCharacterReference(cp) {
- this.charRefCode = 0;
- if (cp === unicode_js_1.CODE_POINTS.LATIN_SMALL_X || cp === unicode_js_1.CODE_POINTS.LATIN_CAPITAL_X) {
- this.state = State.HEXADEMICAL_CHARACTER_REFERENCE_START;
- }
- // Inlined decimal character reference start state
- else if (isAsciiDigit(cp)) {
- this.state = State.DECIMAL_CHARACTER_REFERENCE;
- this._stateDecimalCharacterReference(cp);
- }
- else {
- this._err(error_codes_js_1.ERR.absenceOfDigitsInNumericCharacterReference);
- this._flushCodePointConsumedAsCharacterReference(unicode_js_1.CODE_POINTS.AMPERSAND);
- this._flushCodePointConsumedAsCharacterReference(unicode_js_1.CODE_POINTS.NUMBER_SIGN);
- this._reconsumeInState(this.returnState, cp);
- }
- }
- // Hexademical character reference start state
- //------------------------------------------------------------------
- _stateHexademicalCharacterReferenceStart(cp) {
- if (isAsciiHexDigit(cp)) {
- this.state = State.HEXADEMICAL_CHARACTER_REFERENCE;
- this._stateHexademicalCharacterReference(cp);
- }
- else {
- this._err(error_codes_js_1.ERR.absenceOfDigitsInNumericCharacterReference);
- this._flushCodePointConsumedAsCharacterReference(unicode_js_1.CODE_POINTS.AMPERSAND);
- this._flushCodePointConsumedAsCharacterReference(unicode_js_1.CODE_POINTS.NUMBER_SIGN);
- this._unconsume(2);
- this.state = this.returnState;
- }
- }
- // Hexademical character reference state
- //------------------------------------------------------------------
- _stateHexademicalCharacterReference(cp) {
- if (isAsciiUpperHexDigit(cp)) {
- this.charRefCode = this.charRefCode * 16 + cp - 0x37;
- }
- else if (isAsciiLowerHexDigit(cp)) {
- this.charRefCode = this.charRefCode * 16 + cp - 0x57;
- }
- else if (isAsciiDigit(cp)) {
- this.charRefCode = this.charRefCode * 16 + cp - 0x30;
- }
- else if (cp === unicode_js_1.CODE_POINTS.SEMICOLON) {
- this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
- }
- else {
- this._err(error_codes_js_1.ERR.missingSemicolonAfterCharacterReference);
- this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
- this._stateNumericCharacterReferenceEnd(cp);
- }
- }
- // Decimal character reference state
- //------------------------------------------------------------------
- _stateDecimalCharacterReference(cp) {
- if (isAsciiDigit(cp)) {
- this.charRefCode = this.charRefCode * 10 + cp - 0x30;
- }
- else if (cp === unicode_js_1.CODE_POINTS.SEMICOLON) {
- this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
- }
- else {
- this._err(error_codes_js_1.ERR.missingSemicolonAfterCharacterReference);
- this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
- this._stateNumericCharacterReferenceEnd(cp);
- }
- }
- // Numeric character reference end state
- //------------------------------------------------------------------
- _stateNumericCharacterReferenceEnd(cp) {
- if (this.charRefCode === unicode_js_1.CODE_POINTS.NULL) {
- this._err(error_codes_js_1.ERR.nullCharacterReference);
- this.charRefCode = unicode_js_1.CODE_POINTS.REPLACEMENT_CHARACTER;
- }
- else if (this.charRefCode > 1114111) {
- this._err(error_codes_js_1.ERR.characterReferenceOutsideUnicodeRange);
- this.charRefCode = unicode_js_1.CODE_POINTS.REPLACEMENT_CHARACTER;
- }
- else if ((0, unicode_js_1.isSurrogate)(this.charRefCode)) {
- this._err(error_codes_js_1.ERR.surrogateCharacterReference);
- this.charRefCode = unicode_js_1.CODE_POINTS.REPLACEMENT_CHARACTER;
- }
- else if ((0, unicode_js_1.isUndefinedCodePoint)(this.charRefCode)) {
- this._err(error_codes_js_1.ERR.noncharacterCharacterReference);
- }
- else if ((0, unicode_js_1.isControlCodePoint)(this.charRefCode) || this.charRefCode === unicode_js_1.CODE_POINTS.CARRIAGE_RETURN) {
- this._err(error_codes_js_1.ERR.controlCharacterReference);
- const replacement = C1_CONTROLS_REFERENCE_REPLACEMENTS.get(this.charRefCode);
- if (replacement !== undefined) {
- this.charRefCode = replacement;
- }
- }
- this._flushCodePointConsumedAsCharacterReference(this.charRefCode);
- this._reconsumeInState(this.returnState, cp);
- }
- }
- exports.Tokenizer = Tokenizer;
- //# sourceMappingURL=index.js.map
|